Carotid endarterectomy can be complicated by neurological events due to different mechanisms. Monitoring cerebral function is difficult under general anaesthesia. By contrast, the monitoring of awareness and neurological deficit is very easy under regional anaesthesia. The aim of this study was to assess the requirement for arterial shunting during endarterectomy performed under cervical epidural anaesthesia, to analyse the neurological events and to compare the value of clinical and electroencephalographic monitoring. Sixty-four patients were included in this study. Cervical epidural anesthesia was performed with 0.375% bupivacaine and 100-150 micrograms fentanyl. In 19 high risk patients, a one-channel electroencephalographic filter processor (Cerebral function monitor, Critikon) was placed over the affected hemisphere. Before surgery, it was estimated, on angiographic and Doppler data, that 18 patients needed an arterial shunt for carotid clamping. An arterial shunt was in fact inserted during surgery in only four patients, because of cerebral ischaemia after carotid clamping. Transient obnubilation was observed during carotid clamping in three other patients; it disappeared on unclamping in two, and on increasing blood pressure in the third. An additional patient suffered from a very short loss of awareness after declamping, due to cerebral embolism. The cerebral function monitor never documented false positive results, but failed to detect one out of every five neurological events. Controlateral carotid occlusion and preoperative stroke were documented to correlate with peroperative neurological events. Cervical epidural anaesthesia, which maintains consciousness during surgery, reduced in high-risk patients the need for arterial shunting as well as that for the analysis of neurological events.